For the past three years, the artificial intelligence industry has been racing to build the most powerful underlying models — the GPTs, the Claudes, the Geminis, the open-source rivals jostling beneath them. Each new release has been benchmarked, ranked, and treated as the latest move in a high-stakes contest over who has the smartest machine.
That race isn’t over. But quietly, behind the headlines, a different and arguably more important race has begun. It is no longer mainly about who can build the smartest AI. It is about who can actually get those AIs deployed inside large, complex, real-world organisations — and turn the impressive demos into measurable business results.
On May 11, 2026, OpenAI fired one of the loudest opening shots in that second race. It launched a new $4 billion company designed to do exactly that.
What DeployCo actually is
The new venture is officially called the OpenAI Deployment Company. Internally, almost everyone refers to it by its nickname: DeployCo.
It is a standalone business unit, majority-owned by OpenAI but structured as a separate entity, capitalised with more than $4 billion from a consortium of nineteen outside investors. The lead investor is the private equity firm TPG, joined by some of the most powerful names in capital and consulting — Advent International, Bain Capital, Brookfield, Goldman Sachs, McKinsey, Bain & Company, Capgemini, BBVA, SoftBank, Warburg Pincus and others.
That investor list is itself part of the story. The presence of Goldman Sachs and BBVA — two of the world’s most influential financial institutions — alongside McKinsey, Bain, and Capgemini, signals that DeployCo isn’t just an investment vehicle. It’s a partnership between the company building frontier AI and the firms whose clients are most expected to buy it. The banks invested. The consultancies invested. And those same banks and consultancies are also expected to be among DeployCo’s earliest customers.
The mission is more specific than “sell AI.” DeployCo’s job is to embed specialists — called Forward Deployed Engineers, or FDEs — directly inside client organisations, study how the business actually works, and then build custom AI systems tailored to its specific workflows.
Why this matters more than another product launch
To understand the move, you have to understand a problem that’s been quietly haunting enterprise AI for the past two years.
The frontier AI models are genuinely powerful. A modern model can write code, analyse legal contracts, summarise medical literature, draft customer service responses, and dozens of other tasks at a level that, three years ago, would have looked like science fiction. But a model is not a deployed product. Buying API access to a powerful model and integrating it usefully into a Fortune 500 company’s complex, messy, legacy-laden technology stack are two very different things.
The gap between “AI can do this in a demo” and “AI is reliably doing this inside our company every day” has turned out to be enormous. It is the gap where most enterprise AI projects have, until recently, quietly died.
DeployCo is OpenAI’s attempt to close that gap directly — by sending its own engineers into client organisations rather than waiting for clients to figure it out themselves. The forward-deployed model isn’t new in technology consulting; Palantir built its business on it. What’s new is seeing it deployed at this scale, with this level of capital, by the company that also builds the underlying AI.
Denise Dresser, OpenAI’s Chief Revenue Officer, put the strategy in plain language at the launch. “AI is becoming capable of doing increasingly meaningful work inside organisations. DeployCo is designed to help organisations bridge that gap and turn AI capability into real operational impact.”
How the engineers will work
To staff up immediately, OpenAI announced it would acquire a UK-based applied AI consultancy called Tomoro, which brings roughly 150 experienced engineers and deployment specialists onto DeployCo’s books from day one. The Tomoro team has already done this kind of work for clients including Tesco, Virgin Atlantic, and the video game company Supercell.
What these engineers actually do is hands-on and specific. They move inside a client organisation — sometimes for months at a stretch — and study how the company operates. They map its workflows, its data, its existing software, the points where humans are doing repetitive or error-prone work, the places where decisions get stuck. Then they design and build custom AI systems that fit those particular workflows, integrate with the client’s existing technology stack, and start delivering measurable results.
The metaphor used in some of the coverage is medical: think of an FDE less as a salesperson selling a generic prescription and more as a specialist running a thorough diagnosis before recommending a treatment plan built around the specific patient.
This is, intentionally, the same territory that the world’s largest management consultancies have occupied for decades. The difference is that the people walking into the client’s office now don’t just have PowerPoint decks and process diagrams. They have direct access to frontier AI models built by the company they work for, and the engineering skill to wire those models into the client’s systems themselves.
Why banks are at the front of the line
The prominence of financial services in DeployCo’s launch is not an accident. Goldman Sachs and BBVA aren’t just investors; they are also strategic priority verticals.
The reason is straightforward. Banking and financial services are industries where the existing workflows — risk analysis, fraud detection, regulatory reporting, document review, customer service, trading research — are heavily based on processing large volumes of language and structured data. They are also industries with the budgets to pay for serious deployment work, and the regulatory pressure to do it carefully rather than cheaply.
A bank that successfully embeds modern AI into its operations might be able to do significantly more business with the same number of employees, or do the same business with fewer. Either outcome is enormously valuable. It’s also enormously difficult to achieve — which is exactly why DeployCo’s pitch lands hardest there.
What this signals about the industry
DeployCo is the clearest signal yet that the AI industry’s centre of gravity is shifting. The first phase of the boom was about who could build the most capable model. The second phase, which DeployCo announces openly, is about who can actually get those models deployed inside the largest organisations on Earth — and convert their capabilities into revenue, productivity, and competitive advantage.
OpenAI is not the only player making this move. Anthropic has reportedly been working on a comparable enterprise deployment effort, and Goldman Sachs is the one investor backing both initiatives. The race to build frontier AI is now, increasingly, running in parallel with a quieter and slower race to embed it.
Whether DeployCo delivers on its premise will depend, ultimately, on execution — on whether a few hundred engineers, growing to a few thousand, can actually deliver measurable results inside the most complex organisations in the world faster than competitors can.
The $4 billion in backing suggests the investors believe they can. The next two or three years will tell us whether they were right.
